| Internet-Draft | Secure UAS Transport | September 2020 |
| Moskowitz, et al. | Expires 31 March 2021 | [Page] |
This document provides the mechanisms for secure transport of UAS Network-RemoteID and Command-and-Control messaging. Both HIP and DTLS based methods are described.¶
This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.¶
Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at https://datatracker.ietf.org/drafts/current/.¶
Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress."¶
This Internet-Draft will expire on 31 March 2021.¶
Copyright (c) 2020 IETF Trust and the persons identified as the document authors. All rights reserved.¶
This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (https://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License.¶
This document defines mechanisms to provide secure transport for the ASTM Network Remote ID [F3411-19] (N-RID) and Command and Control (C2) messaging.¶
A secure transport for C2 is critical for UAS Beyond visual line of sight (BVLOS) operations.¶
Two options for secure transport are provided: HIPv2 [RFC7401] and DTLS [DTLS-1.3-draft]. These options are generally defined and their applicability is compared and contrasted. It is up to N-RID and C2 to select which is preferred for their situation.¶
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all capitals, as shown here.¶
The FAA defines the Network Remote ID endpoints as a USS Network Service Provider (NETSP) and the UAS. Both of these are rather nebulous items and what they actually are will impact how communications flow between them.¶
The NETSP may be provided by the same entity serving as the UAS Service Provider (USS). This simplifies a number of aspects of the N-RID communication flow. An Operator is expected to register a mission with the USS. If this is done via the GCS and the GCS is the source (directly of acting as a gateway), this could set up the secure connection for N-RID. The NETSP is likely to be stable in the network, that is its IP address will not change during a mission. This simplifies maintaining the N-RID communications.¶
The UAS component in N-RID may be either the UA, GCS, or the Operator's Internet connected device (e.g. smartphone or tablet). In all cases, mobility MUST be assumed. That is the IP address of this end of the N-RID communication will change during a mission. The N-RID mechanism MUST support this. the UAS Identity for the secure connection may vary based on the UAS endpoint.¶
Some UA will be equipped with direct Internet access. These UA will also tend to have multiple radios for their Internet access. Thus multi-homing with "make before break" behavior is needed. This is on top of any IP address changes on any of the interfaces while in use.¶
Many UA will lack direct Internet access, but their GCS may be so connected. There are two sources for the GCS for the RID messages, both from the UA. These are UA B-RID messages, or content from C2 messages that the GCS converts to RID message format. In either case, the GCS may be mobile with changing IP addresses. The GCS may be in a fast moving ground device (automobile), so it can have as mobility demanding connection needs as the UA.¶
Many UAS will have no Internet connectivity, but the UA is sending B-RID messages and the Operator has an Internet Connected device that is receiving these B-RID messages. The Operator's device can act as the proxy for these messages, turning them into N-RID messages.¶
The UA MAY use its RID private key if the RID is a HHIT [drip-uas-rid]. It may use some other Identity, based on the NETSP policy.¶
The GCS or Operator smart device may have a copy of the UA credentials and use them in the connection to the NETSP. In this case, they are indistinguishable from the UA as seen from the NETSP. Alternatively, they may use their own credentials with the NETSP which would need some internal mechanism to tie that to the UA.¶
Command and Control (C2) connection is between the UA and GCS. Often this over a direct link radio. Some times, particularly for BVLOS, it is via Internet connections. In either case C2 SHOULD be secure from eavesdroppers and tampering. For design and implementation consistency it is best to treat the direct link as a local link Internet connection and use constrained networking compression standards.¶
Both the UA and GCS need to be treated as fully mobile in the IP networking sense. Either one can have its IP address change and both could change at the same time (the double jump problem). It is preferable to use a peer-to-peer (P2P) secure technology like HIPv2 [RFC7401].¶
The raw RID and C2 messages will be wrapped in UDP. These UDP packets will either be transported in ESP for the HPv2 approach or DTLS application messages for DTLS. In both cases header compression technologies SHOULD be used and negotiated based on policy.¶
For IPv6 over both WiFi and Bluetooth (or any other radio link), Robust Header Compression (ROHC) [RFC5795] and/or Generic Header Compression (6LoWAN-HGC) [RFC7400] can significantly reduce the per packet transmission cost of IPv6. For Bluetooth, there is also IPv6 over Bluetooth LE [RFC7668] for more guidance.¶
Local link (direct radio) C2 security is possible with the link's MAC layer security. Both WiFi and Bluetooth link security can provide appropriate security, but this would not provide trustworthy multi-homed security.¶
HIP has already been used for C2 mobility, managing the ongoing connectivity over WiFi at start of mission, switching to LTE once out of WiFi range, and returning to WiFi connectivity at the end of the mission. This functionality is especially important for BVLOS. HHITs are already defined for RID, and need only be added to the GCS via HHIT Registration [hhit-registries] for C2 HIP.¶
When the UA is the UAS endpoint for N-RID, and particularly when HIP is used for C2, HIP for N-RID simplifies protocol use on the UA. The NETSP endpoint may already support HIP if it is also the HHIT Registrar. If the UA lacks any IP ability and the RID HHIT registration was done via the GCS or Operator device, then they may also be set for using HIP for N-RID.¶
Further, double jump and multi-homing support is mandatory for C2 mobility. This is inherent in the HIP design. The HIP address update can be improved with [hip-fast-mobility].¶
DTLS is a good fit for N-RID for any of the possible UAS endpoints. There are challenges in using it for C2. To use DTLS for C2, the GCS will need to be the DTLS server. How does it 'push' commands to the UA? How does it reestablish DTLS security if state is lost? And finally, how is the double jump scenario handled?¶
All the above DTLS for C2 probably have solutions. None of them are inherent in the DTLS design.¶
The cipher choice for either HIP or DTLS depends, in large measure, on the UAS endpoint. If the endpoint is computationally constrained, the cipher computations become important. If any of the links are constrained or expensive, then the over-the-wire cost needs to be minimized. AES-CCM and AES-GCM are the preferred, modern, AEAD ciphers.¶
For ESP with HIP [RFC7402], an additional 8 bytes can be trimmed by using the Implicit IV for ESP option [RFC8750].¶
NIST is working on selecting a new lightweight cipher that may be the best choice for use on a UA. The Keccak Keyak cipher in [new-crypto] is a good "Green Cipher". The Implicit IV, above, can be used as the Unique Value in the Keyak cipher, saving sending the UV in the ESP (or DTLS) datagram.¶
This document specifies the use of DTLS 1.3 for its 0-RTT mobility feature and improved (over 1.2) handshake. DTLS 1.3 is still an IETF draft, so there is little data available to properly contrast it with HIPv2. This section will be based on the current DTLS 1.2. The basic client-server model is unchanged.¶
The use of DTLS vs HIPv2 (both over UDP, HIP in IPsec ESP mode) has own pros and cons. DTLS is currently at version 1.2 and based on TLS 1.2. It is a more common protocol than HIP, with many different implementations available for various platforms and languages.¶
DTLS implements a client-server model, where the client initiates the communication. In HIP, two parties are equal and either can be an Initiator or Responder of the Base Exchange. HIP provides separation between key management (base exchange) and secure transport (for example IPsec ESP tunnel) while both parts are tightly coupled in DTLS.¶
DTLS 1.2 still has quite chatty connection establishment taking 3-5 RTTs and 15 packets. HIP connection establishment requires 4 packets (I1,R1,I2,R2) over 2 RTTs. This is beneficial for constrained environments of UAs. HIPv2 supports cryptoagility with possibility to negotiate cryptography mechanisms during the Base Exchange.¶
Both DTLS and HIP support mobility with a change of IP address. However, in DTLS only client mobility is well supported, while in HIP either party can be mobile. The double-jump problem (simultaneous mobility) is supported in HIP with a help of Rendezvous Server (RVS) [RFC8004]. HIP can implement secure mobility with IP source address validation in 2 RTTs, and in 1 RTT with fast mobility extension.¶
One study comparing DTLS and IPsec-ESP performance concluded that DTLS is recommended for memory-constrained applications while IPSec-ESP for battery power-constrained [Vignesh].¶
Designing secure transports is challenging. Where possible, existing technologies SHOULD be used. Both ESP and DTLS have stood "the test of time" against many attack scenarios. Their use here for N-RID and C2 do not represent new uses, but rather variants on existing depoyments.¶
The same can be said for both key establishment, using HIPv2 and DTLS, and the actual cipher choice for per packet encryption and authentication. N-RID and C2 do not present new challenges, rather new opportunities to provide communications security using well researched technologies.¶
Stuart Card and Adam Wiethuechter provivded information on their use of HIP for C2 at the Syracuse NY UAS test corridor. This, in large measure, was the impetus to develop this document.¶